Intelligent Acidity Estimation in Olive Oil Using an Electronic Nose and Artificial Intelligence
Year: 2026
Type: conference-paper
Abstract: The monitoring of the quality of olive oil is necessary not only to preserve the value of the product but also to ensure the confidence of the consumer because the free acidity is one of the most important parameters used to evaluate its quality. Conventional chemical analysis procedures are usually slow and need laboratory facilities and skilled staff, which restricts their application for quick on-site testing. This work presents a smart approach for assessing the acidity of olive oil using an electronic nose and machine learning algorithms. The created system is composed of two Grove Multichannel Gas Sensor modules coupled to a Raspberry Pi that analyzes the volatile chemicals contained in the headspace of 13 oils. Relative sensor responses were derived from paired air and oil readings and analyzed with Ridge regression in a nested Leave-One-Oil-Out validation framework. The predictions were later grouped into two acidity groups using the 0.8% threshold. The experimental findings demonstrated that the model could capture part of the variance of free acidity, with an MAE of 0.280 percentage points and an R2 of 0.457. The obtained classification achieved 69.2% accuracy and an ROC-AUC of 0.833, and permutation testing indicated a non-random link between the sensor responses and acidity. These results show the proof-of-concept potential of the technology for rapid non-destructive preliminary screening of olive-oil acidity.
Institutions: Mohamed I University, Université Moulay Ismail de Meknes
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7223658346